• Hardware Platform: Jetson AGX Xavier
• DeepStream Version: 5.0 GA
• JetPack Version: 4.4
• TensorRT Version: 7.1.3
I’m hoping to identify defects on PCBs using an instance segmentation model following this NVIDIA example:
However, in the example, they deploy their model with TF-TensorRT on an NVIDIA GPU. I would like to do the same process but then deploy the model from Tensorflow onto a Jetson with deepstream. To make sure I understand how this process could work, could I freeze my Unet model and then follow the procedure detailed in this post?:
https://developer.nvidia.com/blog/deploying-models-from-tensorflow-model-zoo-using-deepstream-and-triton-inference-server/
Any other advice is welcome. I am new to this space so implementation details are extremely helpful!
Marco